It all starts with the medical staff. From diagnoses to allergies to vital signs, all patient findings are reported in the Electronic Medical Record (EMR). There is a wealth of data to potentially be explored but for various reasons – complexity, formatting, inconsistency of approach – Health Care Professionals (HCPs) have historically been challenged to analyse this wealth of data in a meaningful way.
More than 70% of all data in the EMR database is stored as text. Details of patient's histories, conditions and procedures are written in reports and notes. To produce real-world evidence, it is essential that the value of this unstructured data is leveraged. Our solution uses Natural Language Processing and machine learning to analyse the entire medical record to make its content available for analytics. During this process texts are also pseudonymized, which means that identifiers that can be linked to a patient (e.g. names, addresses, phone numbers) are masked.
Once the texts have been analyzed, it is combined with the structured data (e.g. diagnoses, procedures and appointments) and then cleaned and organized into a logical structure. After this harmonization process, the data is stored in our clinical Data Warehouse. It is now ready to be searched and insights derived
Health care professionals gain direct access to the harmonized data via our self-service analytics solution. It allows users to easily generate a list of patients meeting research eligibility criteria or to derive insights into a defined population. Users are assisted by smart algorithms for an optimal search experience.
The data journey starts with medical staff, and that is where it ends. The work that has been put into cleaning data and making it accessible for analytics is done with one goal in mind: using real-world evidence to inform best clinical practice.
